Mercurial > hgsubversion

--- a/hgsubversion/replay.py
+++ b/hgsubversion/replay.py
@@ -52,6 +52,17 @@ def updateexternals(ui, meta, current):
             else:
                 current.delete(path)

+
+def _safe_message(msg):
+  if msg:
+      try:
+          msg.decode('utf-8')
+      except UnicodeDecodeError:
+          # ancient svn failed to enforce utf8 encoding
+          return msg.decode('iso-8859-1').encode('utf-8')
+  return msg
+
+
 def convert_rev(ui, meta, svn, r, tbdelta, firstrun):

     editor = meta.editor
@@ -164,10 +175,11 @@ def convert_rev(ui, meta, svn, r, tbdelt
                                       islink=is_link, isexec=is_exec,
                                       copied=copied)

+        message = _safe_message(rev.message)
         meta.mapbranch(extra)
         current_ctx = context.memctx(meta.repo,
                                      parents,
-                                     rev.message or util.default_commit_msg(ui),
+                                     message or util.default_commit_msg(ui),
                                      files.keys(),
                                      filectxfn,
                                      meta.authors[rev.author],
@@ -203,7 +215,7 @@ def convert_rev(ui, meta, svn, r, tbdelt

         current_ctx = context.memctx(meta.repo,
                                      (ha, node.nullid),
-                                     rev.message or ' ',
+                                     _safe_message(rev.message) or ' ',
                                      [],
                                      del_all_files,
                                      meta.authors[rev.author],
new file mode 100755
--- /dev/null
+++ b/tests/fixtures/invalid_utf8.sh
@@ -0,0 +1,34 @@
+#!/bin/bash
+#-*- coding: utf-8 -*-
+#
+# Generate invalid_utf8.svndump
+#
+
+#check svnadmin version, must be >= 1.7
+SVNVERSION=$(svnadmin --version | head -n 1 | cut -d \  -f 3)
+if [[ "$SVNVERSION" < '1.7' ]] ; then
+    echo "You MUST have svn 1.7 or above to use this script"
+    exit 1
+fi
+
+set -x
+
+TMPDIR=$(mktemp -d)
+WD=$(pwd)
+
+cd $TMPDIR
+
+svnadmin create failrepo
+svn co file://$PWD/failrepo fail
+(
+   cd fail
+   touch A
+   svn add A
+   svn ci -m blabargrod
+)
+svnadmin --pre-1.6-compatible create invalid_utf8
+svnadmin dump failrepo | \
+    sed "s/blabargrod/$(echo blåbærgrød | iconv -f utf-8 -t latin1)/g" | \
+    svnadmin load --bypass-prop-validation invalid_utf8
+
+tar cz -C invalid_utf8 -f "$WD"/invalid_utf8.tar.gz .
new file mode 100644
index 0000000000000000000000000000000000000000..537021f18693380b7f1133bc0d8b2558a9966706
GIT binary patch
literal 8227
zc$@(vAl%;{iwFQ*K(tQ)1MFRUbK5wQ&)@nKI5{=4Q=;T&a_5xy*5ha<zKQLW?PT^g
zo12A5NamU%bOb3yxw-r7*AIXXNsb?h<;hyCsfkPy4WQBJ{&hD1w%Skb{aT0X#m)}?
zyR*HsMSpK>Z_~f_da|(rZ>(>xZ*0N+jg3usUF_U{)1ExZOezy0o&?6fh*UUG$yM*E
z#4Msgi@jRy)6V{@qfRURl-^ezIKa<&yyuSp*7nZS_-~@Pi1qtwQ2Bb~<9~h;8Ug=F
zah7#Q(8W;4B2hza!c-@dmje1N(hDUTW)#a{5XKFWY8}D<g!&}mZ&eG^mJqMBh;=Gt
z5Hv-gB9*G9(1{r6AnZ*<FN~BC7b;OAi{Ymc6P=06SdYbp9@l`F9O*C+zD|;Cn1*qm
zDlHn)s@2|y@Kao*>2SB*HhF7Xawz=^)zV47EqYWX)Dp(<b0AZEq4v;B{sUjF_J!7$
z=3cG9?e+iWHr!h3{{>ub{D+D1BJE$c(!ntL#&uky|F@s7Z%&Q>4m@A#|3zFY|I+S;
zaob$fYAfRHX~%nWa`20IeRA?kZ3XU}gA#V4Ay87#R)q0LU#dWKRS#9IjKLE4QD(xC
z67*tdps1fHnW_LxL)o%qu>s8vM2J{-!o*PDOQ=Zq6FCr2c-8nxI82*BS@~HiyAhZZ
zXsApka4+>+Ya*5dXhXxbt|3-Y`HwF`{{je59cXpfxTOlj8k)1#6yq?xAmh^0QKZLM
zwN3)COo5O|KO3kxHM`V#AwFz;6sMiHCuiQ<{qxswV-*VyB~s8_M`5DYxX9*5VPEE}
zC=$j~Moq}B&Qc`RWzMO<650AF-ko($t3*%)nXS$~Vi{Xv`=dBGIXdbbpO;GbdVu{A
zx;Op$&PQ=}-r0Y{;|;HGh>%~RTA*WTGNEd;FsYJJP5}5H0VY^qBOV=cAsS6Le$c03
zmIB-339tzH8v~61Za3*=eh#-1d^irH2ncnlW*7sL*93K6s+ep#v6}Abp-R}S3#=Sr
zgb%eNO3_FmAA4X^mVngY8v~f-qMyk`#;IaFL$r<Ps~G9g;3GB(|2qST_7iOkQK}4=
z0-zyM&=IYCas;X;vnUl~osbj<VWNC`nVYRhz<p{lay5&g1b`+$Nk-cm9N)}^&Z0mV
z1p^i<5dC2k`e7PPaA*Mf_|Qbtx{~ecQwU5a@iv+71ITI?tc0rVgnTKwSqdbWW~oq5
zh7oW~tj0R_KC48pmC8%s^#Eix0Wv^A#sLV?tIp}EposwdhK}*f0NIN|5WN=am6I5z
z;JY~dB+LYaX~1-21ZWX9-&X-vj6$cyg#<NHOlxc@=^LS;P~v(^uLjzV5RkzEK)O2X
zUtq@ua$=tY1$+oMjf#4RUYUo-Iz>^CX<3Poq;yqD@1T-$^J3MkfrimeW^Kcuw$*ep
z41F0z#k+)51J-!b056eUR2@CVihu<5me4VA#*5TQ0Z&vi2n}gWgvK<OGk|!-arW0R
z4)oaM%teW}vV=-<MpB?$-s4tRrVRw@#U1E+qlq&TRTC8LN)!4b43wl1f;&dKDezkH
zxRDKzpkr-KPJQDfutRU{3@|NK2~-CE->ub1yYAJWZq#e2HR0E0y@sj}er?rjR;9ro
z+w~eLEcm%oukjR+ESn(1hSW`fP0?c%LKrj&ld;!>27&?vh8i7%XQJZ<I4P+3Vc2l~
zk?u9AjzH6_s#d>cEd=^IOy2XL7rqj$R;#f_^1ON^CtbyQhI<N{mZvVXK@DIcxmmzE
z3!0St`&#Yn{qY;Htb6UuB+QOc909v(KLZ=~=uRI-xOUcg#eac9tkqCM|90~0G4SN?
zPb~?iTYpOItHTduUS0ey{#FyE>nRczjsH5q2JDIkyzmr=L+9G|R=eH!yXEFsfi_HG
zT7r}r)M>p$gHqxq(v(e9!{Bo^1I)#evL<mJ4pqEc1Lg7oT2(Z?F7`ydE<V1*0x>ia
z6^~f2^}<@A&UloF9O_bV-@vge>_LMEH@QpIPf@0>tp5Z%Hx3P$i?M3JjO?Q;8*_C6
zrXn>3Rkl@gPFe2O6j&co?<9!^ZqhJ7>p&v~6Kl~?6rcW2(VYKn)y2l2e%P#03pO6u
zId%9SQE=)ZUI1_Y8~Xs);QyU1!2DDA|HbC>CH`N;Rl)yf=liGU9_rk~^QsR}7LP0a
zfHKfeKgA`?LAKIi9WrKsNCVBV-wNFq`~);cb3KKdx(R6U%3UZs3s<@dh#T(TSD1zQ
zGD8)W0-QitN?<{z*s9dxaqdImDIDy--T(3M&EffB=S-|3LnA<NMhzt++X&1fGTfI#
z*$tyG4HdV=;<$O21w9PnUV>pmyasrwtAGz{IeywD>8LaSYaO6<M|Q?)j3Zwf*^Sl?
zLm~KqULsXs2vLawL~-k&-y%}dFb=81-0Uk1?A42|AP%zT14S@#FtOZMvRy2G#0r$l
zDvJ>ApzYr+RkCaJKzT&XE2^vW7?BiI@cdfg6SztcZhDT-RgQwRpJR)0;#2{fMpQD8
z04oiMbOKcZdNdi1zJ@sgU<_z!?oDMtvDPMnYD)^pAYCeKajs)_qM{JyO3r4$oZzq&
zokZdjf~G~Mu5xgkD{rS{!(8d`03o9n;qX{B<!hKkai+|zJswK{hcMNEDx8W#pXX{N
z*Sw)al;if>c|;ZcwGgkUsxcpnXH-E#xp!2#dt(nt*vV57+kzp1R)(@BVGM&9GBFj8
z!bHb5ZiBo8aRog^ZVX#b&>lZQqpZi(v^Er20?%Pw2G|N4e!!?6nxit+kxW7wzm#(}
z=m04SLt#~!$v|*MGRjmQqky3?SvME5o^@~{nZ;m0ZAQnzfl(-h>>3?qDce7*qJjAh
zkZBlK)FW98(;YMk6tJzflTuc=)wG@PvuK!D0me{K%*|ZY1kVjnT$|fRV@KiZ991Op
zlS?hx^6YS;*F1)XrC?wPNjJ$>!i+N*`mqt7xw$p4ZYbt9pEar^vgob0G`&)ytDVA(
zxol(h%@At$xbdcqEx8LR)R0NotrDwMoO0`mgHv8(;rGb*0+tLT@Qh2n$(>QpqiO71
zuHSzb*3&UC-l2><1UMcv&%bO9C)2Nc{FrC^R;e^}a33%NmBR^P3<uFl&Y+#1HS60w
zG`9X8t~~#v&3(Cl*ZF^2&tGgz`+qyn*O&g^BCcxw=j3d@`&W{V8ShW_5bIy<#<_s}
z7ZN_#yKz9o<$H>z+{lY7d^n(Se{S4q=e?-9ad9^20%L6IB~hc0Sn>~qQQZEN(mfBG
zNcd6t(k=0D11wD*!h%~vT+0!VvGZkY@dDHil$*9*(zm`2l<RE;@1U&o0oqb7a2gGv
zX$PBJUS)#e#)imJ>xuyWRnd@<JFPddhyYy6XyWCZ6ltQ=oxycksuq)RD2wr9f!1Xs
zQ{){=H<!SA39KIpum*!+aW$rk5Km8>0xqR>;H|o7<5)xU+V)ML-uqVIw!Hr5p5Whg
z`0x47_RRX9t>+s{{I`f}X8q4x_?HXEEn@$uTbyye{UV@SuK}8cgb%$6==?9o-tqoX
zrvwSnFj|O&Oe!z`nSz8s;_Up@;qga-nM3dV<d@FznRxbWYH^IAkT*NSWFM$d%aQ=w
zKz~yxREDulr6Pu^m&oq^Ew*9pI{pe2M*4v=(Lf6T{6n+_<=m1G_-RV41Hr&{Ot%e&
zf@lzna{u6*Y;lk;O0v=NflMx4;+~pl8x8b{h>jzCWE&#<3z<ltO$a^j;NqOpvLQ#q
z(Bc4%s|DVj9(v075+k}ed`;U0n+36$%Vq^YjPg1}<M&4KvU{;q8Lwq(5}ERAd3(RB
z6+<qr?L_j{0dz~iE)~M8ge8XUmzN+rFb{oo!-%;k+y*G`8h;nJ0ULIAAED4P_va!A
znB&icI6gV=?27Z(9a6-gm`{E_-9HjX@6OKgnK(Lmb@<bN(hHzwnxgZUgU;J?DqIlx
z>Esms(!;aElVc?DwsXoU!YIV_(kR}IG^#!NiBUJ5thIzCLS)?Tfbqoymxq9$(t<+F
zN)}7>W#G}=oEd}`D<RHR^)g<LSF!`h4Hfg^8X}cN);%g%O(Y1Slrm&)R{<8X#6V~Q
znMz7Ce~Jz5J)m`pKt>UfZ0KO{K*7Y0Lo>kPqEs6UjgdWNhtFuSN=6B{5ZHba2|FS1
z7K^6SVMwp9407)^O#KnAQAxu71+JzUg9->yN{B>JH*yx?E}9w}$)w4NpcneYfdY~|
zYan1Q_p)g2+O<dLNET?dvVQHhZsgL7TzZk;wHFD>-+GZN;)oUvWQkP_j8OWZag33D
zRbW6|*8ouXQ`)+EL1Jm0VPAOYl0r@O!ZWoC4RN)isDF^&=Y8+5>)yY8tcyR1cA!S>
zIE$ic^5y^&H$~If-mn7Au?EE+NmHQmBy$SN1l6z|Df^mT|G>M-aGV{8!i5lBMT-Z~
zM1<ga<asO7Lmn|zN0`1U`PtOF`7ipt{y%XRs*?Y?rwu^Y<$pG|pHKV$n;Xmhp9{HW
z^FMPo06E#ny?;<odESS&wivtGbusy$=?y;*o&C9DuMUDOyErI!7>kS#QZ=j)$7uct
z@YyxPC9mH4Ld3q@!xyU(84KF8vb1iq$!;<_gKk_bnIW!M%N@(6!$#;(MKZRI4!(<;
z%r%3XC{f}11UKfPF$>nf>^n;f8WcE;frH7~gFC+E3>OP=DZkZK&|Eh}f7BG0Z60ao
zuS#$&a6?ToOCOC-l3UPfYU_4;YQ7!}?_RZjXGNS<wx>`TWeDQSgcr(2irP2et@(C;
z%~<AZ9Cu6Eis>yVyng3ClNHPBn_uPpPohR3&qI%+8Q%^FUc3KeCjavS3M}#8BCeVI
z&uQm9h6=rd*ZapmcV_cIRUx@LACxb6!M9bCPEflWJRbxNy7SIZP4a0WH_j2&R_?F?
zgu_GCwxcDy0fhWgt_a(5ZwrWW4zfd=j+MN@2Jv!^$?m74k}h8BhNIXvj>BEeQ@M9+
z&5W`t%b1E?<ogHb7+w~0N{7xDZ|paC*c(R;JMa{sZM<sm*iYxYuu1Rt-w2oR!fhlu
zS&Ds?knL33>o$oFr0E2iwh!KZuRw_e#tU(KDU)i=3oc#@1k1?^AV6DNk%ipp>BVG@
zC*1R2_fL-xkAL39h+mlFI~(>Y<`N|lh%g}uM~qV|NDhN2Hs(wn3wW!2&ec3FN<=Cl
zeT0U%lc%c)h5>L0O9jczfr)5fzOV;^T+Kl<Knz3e$aj{6)<2r=Ft!j~q<6>fJExf5
z=)3}*qGZfcPLVt5hPszU<xv8f0id2F^V4voUJ7cS24t7ZRUdFaw*n*i(70dAsgJxj
zT?zOn8%~wS=-L>!_Iy>IX;xD$W)u8Wy^jaH&%&QdVqxmpGQqG+Fg%U~L%vL{IFrHQ
z1wNF4?Y}*h!S221d$jNVs4hH-=M&gMbknAu^X>R{GyY3~yYYEL{Q*uil#_X?pOqkp
zWt?ZR5e_8!BJ74BQFZZ$*<b%hzbf&6n8^2$0k|&yzrOWiX8*^|i>3d!h-)_f|L*vT
z3;;_9t4%iJQyBn$VNp2%uI)R?0ubWVlbh!O951iU1fUZu%~xguTy%X-CLnLf<H!XR
z1NBX^0bdoaw-Ua%fW4bij|TOZj9xWpZ*PA0Vf#c4^r#xPw`zPDxL;mRs`3AOf`8YZ
z|GKd;yZ-}TTjIY3Tr=_ix$rL+j$57qn_JUGoB}%w2y5TtIj|0JU3(I&W$yg5U}<MZ
zrA%qJ0#}p|P7%!i@wBu5ixYv_BVoHqDUmRH7VHB8ySto+n)m<KF@HN39^y>Yg<`_e
zq<S=vu&{%b0Pt-A;Wx&E^Ug-4Db7JSnzwUKM!nr}s5rje1T_RA+PfZd`cHPssXg!0
zYLKry*_5T|%V(S3tinZ}XZk&yW4gTVc2(lP_p|`$I{c5Dy{FdyY_2cYe=p>kjsJde
z0nn_JTzMYQyahlQZlf^UotFVQbaD4{VCVGVMnUAlW4PBK@)y<tafldq$UMMOpuBRs
zCgesdfbjjVUi^b=D{i{@rySs|EbjSUjx?I;>DL92A7mBDyy^95yoWhfv>I5x>q3&{
zfk(@gB#&t&Ns+H8)cEGi$w<x(Dz7gomY*y$8rNSf9J=29egyUo-~ZcO-&}tGZxNRr
zf0*gezmcLj@~!K*?)<Oy&FTEl_KVHs_x~1ht*or@DKF&l*gfxzJWv$Ee#T#0b1>Km
ziU_0{U<ffJb6i6@h<HP?{v1(s2y4dM0o1W-zum^CI^pX?A_rv&flsnSXp`DtGO>lb
z;@#<+;)4!_r$(j2BmtLVB;&NTR--$2|7L9M1AiJW`dDKgVUkP7_H+QU*mO{}{af4b
zGguB*`mTsV$y{TNSzP_j$Es{ywq%PcnGAFWn3zP4Ei$rZ=CdFMgN1mgHZEk35W8ZJ
zWfa=A=)6Y0P6x`=o1%{XPn{8S9coc^E^(&P{DXq*UO}t*g?vpVxb5IG&|yF1c>@U~
z3%_3y4W`WlX-a1TNt<_}I=Lb4CFh3;%N-fC>4hrms@U8PgkjiaxZ#FBN|K*IW>Zi+
zMs((Q&$_QUZ=JXTd&yOpZt!g7O^gmiQy*uUM<(aZ7dujsr9CQGTd4@t9uvv?jQ%e8
zUV#^Mt*}*RYvht@&f&N369GIL3BMddvjt|~yW$(njzM-AAARS5Kg&o6y4@rKlqZF<
z4O3J^Q#z6;e6Ej0wy%76+lo!YRuu|0g&yKDb^B*;kf4eYYQt<m=*!=lX_n?W?JdFc
zuk6<%x%|7>edm%Gf&Mg01b-2-EEG9gT=f;c?(ttiG2>xCKHrq%e4<JCnQpRBXOUw6
z33==Bw_EJ^y5mK@Fc?H*c0g`A7*6rb8<c{l#b-1*8I(66&b1D5fX;BcvKN$2)v>z$
zcvP5{4@%DWJ^oqC&6FJoZe?f_D5Xisj**N8#a^kZ{=kIHSQodE2A{HJo$a~Qi0PC3
z%Q`D7Q)P0oax;xp-YE3>gfeO4M6TOJlx<eoFv5stS%_k7JFzE@CgP_sF{wDs-^#_4
zj!yGjsL%6phC4~cFqkRKHGVb<9(x=#3O}n{cVSWE)^VUxmWj0n10nl(Vo*9O-d41E
zB|}9}av(ud9VMl!M<`{phAA5kRe~o^f>_{Bj|fpO${|dQ^3>^@6Z|gyocoqDmDYHz
z^0%^v^kvt)r_y*K2MBP@1<`@lT42z03rBcl%92W3+Cn@ckN1YnaDGo<wE!H9G&yRR
zC$i%UMe#SBbYmH(<#};1Nisv~kO)L!H<1=GROPZIUfG7>M3#|Jt)fb!cxi&&i|2xs
z2-b3m2svQX%|cwRfy#yIW=MeQ*ubFZBz;M;gXbMJ>D-q_PQmjm*BT`38056k51jM4
zS*IPKMZc!!7lu+s;Gm=8Z()oQSm7ZHfuU)V<rXlPZBsJpw2CFfu9-AZaUXi&fU5g?
zRwO5Q_TGw(KIKfOw%|(VM32G%By^p25K`(_#eOmw^77~nzRe;Lr>c)s`ST_GZQ}^Q
zffegWCVjUvWrKv=*9qxm3l^+uR=pmhHgO4BCE|u7>xzo`W29nZ^IvVV;Fs6M0}+8t
z@!DF8G=Z&uJ`h5I9fYx0BEBcq@uvJ#y}7ybd>J5lRG0PtNqgT{9oP8(o9i#8<3C%Q
z%l$u#xvKtuHLN|0%}7V(u_DG(DIr{Nn<*<6ux7VQgCw1hI?%yyBw)YcH<f2$G9_x5
z+CClgI*zR-qYth-W5x>wO_e~No?!RIA%Imj6*5{Y``yk22gOm(OgiQyO9}N?DnN&|
zCHAM<Zcmt&2oZaz_4O{15FYM5=VP-`YMRB}exkD>vV@&RyYvxZloa;sSmFb{WInpa
z6~*DLD3n(fqi1SfIlusOg9pVSzKy4q7aF`yG^fCxv2&j17IzP)i^_O{S=I8(&l-qJ
z0$(QVzwyBHiYQ`dmllaC_tnfHus8gapEsP5ABbP*{=Y^`oDvT8t+&!o(6W;ozS>9{
zYkB;E9DMf5;la4{1f(Z+$v_sJ27To_=O@_7e`rK^?Ss9){Iz{{zV`}Vh1Rv-?48AU
z=H>wag7%61<0Ool;!PT~+9&V!P64&Z1iEzyC^}5DRJ9L|_x=OO?L+@$AJL*6ZpvCF
z$>cW>;ol4;_4fEqlm5z|(_i!!H>>nc^9Q(v3pqZtds?@<GRetekYNme{J@3qFyd!o
zkKQdN$SxH5-0o$cvDlwyn?n?yQSpb?Tg&U)T-N?`+~c0=xW@jkZ$ICew*T8(%l#h<
zxxQroE3vX`O}_~T;KU_m!(YiNwF={23LCh(*xuW(_R=xMxa8m2XK)TxgmJ<DZ|~Z6
z+_r%z`l_!G(3dudY?8Wqi##l_8ze}u36f0<q$mnXq<7WYmMh8G><72s(8qFy)WvpE
z*FmJDyF5u?#21p{a43pAXXa-#!!O;v8mSm~eBL$|9;0C0fP@fz^L~h>ffRM4+0U-d
z$>5msdb1?Hs`))DB^;_ZQy;!IZ+~f{eMU^C+JZyYfj<5({;wVZcF%u<(8hlo1>s#9
z%oayuG8O5t><{#Rb{zkM!g9g#KSluBzka{+vP}uZKnz#=nd>;T{@b^I5Xa2U|6`!F
z|GF${2zXb0{@LJo{|`bMS^pmcXn*7TL)CHE|997aKJ)GV?<lakz55OHf8ZPpSjR#C
z4|ex|Sm@jP|0oE;l!qdao=k=B1>B!`X-cKXxCrNr2SGTG?G$l$f&OnqNdqkX5XZ0o
z{E*q_zfrJle=^7j_yqe$k#Frk26pXF2dLxF{x1!8^S}9VVD%rP;F(K_)8+qleuV#b
zec?QF6Z}~4TB;9O>@c-F^GN8?_>zzl;-C1l35jMwlDt*ljvW(`S(lV9{!ITWugbrL
zbAEAV7ZP6`cKv?{pF4Q|6YV|!$M*Su9Q+smPLzKWnnZ7{58pYU|Ie#=HHh+4<G}O3
z9|gPd-~Ax6_y19Ftp4-y>&LHqku`uqv6RaU)Re+QvIo*B_9*+v(`dixns1_rOKVdy
z3=LU>pkjfww8Axr2qa}A;=H-JnFzkR;%dMyr)`NJmT|g<VcLETWH?%E(X|E1gixc_
zHvm(lDXRUFmSX8fKim=)BGp89&Jyw+L6o(%p@%{~ZXkuK4lAqLbdUSUr7WHh^@pBN
z$|yy(&$6hc*EQ;qyS(jMofYMJVyH53-S|7!l3+}Q*vDOHnjd<K&b(?22z}!|dGY%4
z<mHQ3my_-8Y3*WYnAJ5eG01Cu15ulnn?;=iTc@v{(gM)y;s%l>HF{tjFD9>a*pvql
zmgRvfS6z1UJ^huDszJ7^tajhLd3t8TVnEn=q-kQPQPWNs7dmnun5*2DMlB(Ecq0+S
z(1$A>@P=m@5|n|<qecQot>KHx5SF@LxaOJ)3dGQDv5HQZ#YF`TLOyAQQS{&{G008-
z8r7TSNBS2e8a=&f-bBPHH&si}1>eG-8Hf4_G|yc3fM-M{u7p5+Zjogh7?@h!rHbu^
z3DKiLjS4Lt!JtRV&GXd;BZn7PS6b0)ipm@L431SGgsow1Smvs_h3?;;xa&J84r_oy
zaw8-{rS^1cF7A8nbW8rbHgd0RD~Xk03(XvWtpIj$PecgWrID}oEe4G7s=8P=>L3UB
zs~!wg)dDl~fb7d?l&Up{Yz2*EQqNjAyc>o%7~x%!gMtXJLJ!0Z#F|x?3}Z_tEYoao
z)bZ(j^%AooaJO3Ec*}~a#yBnN=O$@A$4m)oQdsi5RbKK=jSCbx?O{8$w1j5m43HJA
zO?+Rg8&qYqgm`Igg+Xxih!S!wS1`cc&DtdFs~V}_o*dEyk=iU%WdikSk^UyLy4vF4
zag*<iFOfJ0L4<mpqjMALsR5?HhdSa9bf`O5_Uqp_(o%KCwb70nrzmmcHIax~D){AA
zOy)H&)wBj7<+L^!<&urP_d_XzoV9Md52i3RBGL6!Ti|kv`5ep!$~cpPP7V50R<-U=
zPxq$@4yvbtE~!N$Pln;8rYhCbkYeMVERE82E<Ywa-D-5Fb5=fYR9Y~&^}AmGyJnOR
zcK^rqA5LULFh4Yo=YRN-<^PU@=J|i9y??;*`@bw=cK>%2boL)+?;r39_KzbQ|7jHT
z?LWlcKj6^*Ur6`#A0)8n|D)jS{Q2c6glzxj{Nn7@`Aa7kCob`*dO9uXl*UedyOt-o
zc6Lxbp+0qL`Ch$Jm$-rYCju&xOw5uwAt@t)oF#<JW{J$^Nx)+DJ9vgWai^!Mg!OG%
z6i|peb)1*aPoG{m&s^%9!f`?5;3h=BcGz(2H>~-9n0AT0bkxeZ`G~-8OfzZQOv*yn
zXYiLt^C<HL7YR)`nFpar7|SFP5%CiuMBCKemi`LG*Tf?bxfqvZI_3ggl;f!OxNP>n
zq|*2Oe<*)Dc>Wjp`~FX?|BnIff**1-(s@64*^cuQbpF6)g3aPs&}<gXgCypu3^<jH
zFd>Ma2A%Vxo4Ax%HqloFv;Xgbp@*vD@bll^{htQ*{BIP%(U0}W`-@%u&k%JS_W!;3
zuOzVg&rx9c|Mwf(|4Y8AbT+#o_5qLI{|Wri@_$ExUjGdn0^lJ3i|z40Nl5MdKL&hf
z!EYfDCLA~kDby1hA@-hdZ!np{vEPuVKcAHU!H<cZ|HnXMf025N;;NbqaR_{9{m<y`
z{$CJAHviu!*oyrG@qio7RF9=YC;sF;5T1Z?R8E{HE!V5AAW<3+Au_2WS^u%YXuLqG
zz|hNep6aA$Kn!xW3T3)b$;(9!#eV6oxR>bse-PJ49Pj@=p8s3_A0F2KEi5c7EG#T6
VEG#T6EG+Im{sF;`WR(E$0018sFo*yE
--- a/tests/test_fetch_command.py
+++ b/tests/test_fetch_command.py
@@ -175,7 +175,8 @@ class TestBasicRepoLayout(test_util.Test
             self.assertEqual(repo[r].hex(), repo2[r].hex())

     def test_path_quoting_stupid(self):
-        self.test_path_quoting(True)
+        repo = self.test_path_quoting(True)
+

     def test_identical_fixtures(self):
         '''ensure that the non_ascii_path_N fixtures are identical'''
@@ -186,6 +187,12 @@ class TestBasicRepoLayout(test_util.Test
         self.assertMultiLineEqual(open(fixturepaths[0]).read(),
                                   open(fixturepaths[1]).read())

+    def test_invalid_message(self):
+        repo = self._load_fixture_and_fetch('invalid_utf8.tar.gz')
+        self.assertEqual(repo[0].description().decode('utf8'),
+                         u'bl\xe5b\xe6rgr\xf8d')
+
+
 class TestStupidPull(test_util.TestBase):
     def test_stupid(self):
         repo = self._load_fixture_and_fetch('two_heads.svndump', stupid=True)
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -7,6 +7,7 @@ import shutil
 import stat
 import subprocess
 import sys
+import tarfile
 import tempfile
 import unittest
 import urllib
@@ -301,6 +302,23 @@ class TestBase(unittest.TestCase):
         proc.communicate()
         return path

+    def load_repo_tarball(self, fixture_name):
+        '''Extracts a tarball of an svn repo and returns the svn repo path.'''
+        path = self._makerepopath()
+        assert not os.path.exists(path)
+        os.mkdir(path)
+        tarball = tarfile.open(os.path.join(FIXTURES, fixture_name))
+        # This is probably somewhat fragile, but I'm not sure how to
+        # do better in particular, I think it assumes that the tar
+        # entries are in the right order and that directories appear
+        # before their contents.  This is a valid assummption for sane
+        # tarballs, from what I can tell.  In particular, for a simple
+        # tarball of a svn repo with paths relative to the repo root,
+        # it seems to work
+        for entry in tarball:
+            tarball.extract(entry, path)
+        return path
+
     def fetch(self, repo_path, subdir=None, stupid=False, layout='auto', startrev=0,
               externals=None, noupdate=True, dest=None, rev=None):
         if layout == 'single':
@@ -333,7 +351,12 @@ class TestBase(unittest.TestCase):
         return hg.repository(testui(), self.wc_path)

     def load_and_fetch(self, fixture_name, *args, **opts):
-        repo_path = self.load_svndump(fixture_name)
+        if fixture_name.endswith('.svndump'):
+            repo_path = self.load_svndump(fixture_name)
+        elif fixture_name.endswith('tar.gz'):
+            repo_path = self.load_repo_tarball(fixture_name)
+        else:
+            assert False, 'Unknown fixture type'

         return self.fetch(repo_path, *args, **opts), repo_path